Simulated CBED pattern     TEMStrain    Simulated CBED pattern

 

How to proceed in order to determine strain tensor from multiple patterns?

Strain determination from one CBED pattern is troubled by an ambiguity problem. For the determination of the complete strain tensor (complete set of lattice parameters), it is recomended to use at least three diffraction patterns originating from the same point of the sample. Assuming that a set of patterns in (proper digital format) is given, use the interface in the following way:

  1. enter correct settings using “Settings” command button,
  2. fine tune the orientation (and other parameter if necessary) using kinematically simulated pattern (which can be displayed by clicking on  “Simulate pattern” command button) and the “Fine-tuning”  frame box,
  3. verify what is the impact of dynamical effects by comparing the results of kinematic and dynamic simulations
  4. detect HOLZ lines in the experimental pattern,
  5.  index the detected lines using “Index lines” command button,
  6. save the data for the pattern using “Save data” command button.

When saved, projects are written in .tspr files. These files should be kept in the same directory as all .bmp files containing patterns of the project. Also results of  dynamic simulation will be automatically saved in the same directory. If this directory is changed, the path specified in the .tspr file must be corrected. 

It must stressed again that  the optimization problem solved in determination of all lattice parameters from a single pattern is ill-conditioned (ambiguity problem), and  even if more patterns are used, the results may be erroneous. Moreover, the calculations are influenced by distortions present in experimental patterns. No matter how many patterns are used, one needs to be accurate with input data and very scrupulous in marking the HOLZ lines. Results need to be confirmed by applying a number of computational approaches (with various combinations of patterns, diverse sets of marking lines, changing starting and bounding values of parameters et cetera) and by visual inspection. Even if the resulting parameters lead to apparently perfect fit between the experimental and simulated patterns, one should remain sceptical and double-check each detail.

TEMStrain v.1.3,   Dec. 2014